Details
-
Improvement
-
Status: Closed
-
Trivial
-
Resolution: Duplicate
-
None
-
None
-
None
-
None
-
New
Description
I discussed this possible improvement with Stefan Pohl and Andrzej BiaĆecki at Berlin Buzzwords: By having preset dictionaries (which could be user-provided and/or computed on a per-block basis), decompression could be faster since we would never have to decompress several documents from a block in order to access a single document.
One drawback is that it would require putting some boundaries in the compressed stream, so it would maybe decrease a little the compression ratio. But then if decompression is faster, we could also afford larger blocks, so I think this is worth exploring.
Attachments
Issue Links
- is superceded by
-
LUCENE-5914 More options for stored fields compression
- Closed